Reproduce clinical breakpoints from latest WHONET/AMRIE data

[msberends]

Summary

• Ran both reproduction scripts against the latest WHONET/AMRIE source data, which now includes EUCAST 2026 and CLSI 2025 breakpoints
• Updated data/clinical_breakpoints.rda, data/microorganisms.codes.rda, data/microorganisms.groups.rda, and data-raw/organisms.rds
• Fixed a bug in reproduction_of_microorganisms.groups.R: replaced base ifelse() with dplyr::if_else() to preserve the <mo> S3 class — base ifelse() strips class attributes, causing bind_rows() to fail with strict vctrs type checking in dplyr ≥ 1.1.0
• Added data-raw/_run_reproduction.R: a non-interactive wrapper that overrides View(), sets the UTF-8 locale, and sources both scripts in the correct order (groups script first, then breakpoints)

Flaws encountered and fixed

Issue	Root cause	Fix
bind_rows() type error in microorganisms.groups script	base::ifelse() strips <mo> class → produces character; later bind_rows() with <mo> tibbles fails under vctrs strict typing	Changed to dplyr::if_else() without as.character() wrapper
gsub() UTF-8 pattern error in breakpoints script	System locale was C/POSIX, rejecting multi-byte Unicode in gsub() patterns	Set locale to C.utf8 in the wrapper before sourcing scripts
View() calls abort non-interactive session	View() not meaningful outside RStudio/GUI	Overridden with a print-summary fallback in the wrapper

Test plan

• Verify clinical_breakpoints dimensions increased (now includes EUCAST 2026)
• Spot-check: clinical_breakpoints |> filter(guideline == "EUCAST 2026", ab == "AMC", mo == "B_[ORD]_ENTRBCTR", method == "MIC")
• Confirm microorganisms.codes and microorganisms.groups load without errors after devtools::load_all()

https://claude.ai/code/session_01S5vMqfsiJb59RN2Gyz1tDY